Dr Mike,
This looks like a real move forward. Congrats on sticking with this thing.
We should run a cake stall and get you an 8-core Mac Pro or perhaps a Matlab distributed computing toolbox licence for a biggish cube farm (and a hundred hours on a thousand cores (if your code can be multithreaded).
Regards,
TB
Virtual Polywell
There's another way to interpret this: 10 pixels of realistic particle movement and lots of noise.
If none of you people have ever done software engineering before, I'd like to explain a few things: you test it you test it you test it. And with the most simple things at first.
So in this sense this stuff has 0 credibility in my eyes right now. Either using today's horoscope or MSimon to interpret these images doesn't make that much qualitative difference. And by going out there and beating the drum about this you're discrediting yourself and everyone else posting in this forum.
But all is not lost. So not to be totally negative. Here are couple of suggestions for two very simple tests of your code:
1) magnetic field. Put a ring of current on the x-y plane, and inject an electron on the same plane some way off. What you should see is something like this:
As polywell is made up of rings of current it's a good test. And another thing. We have this wonderful thing called conservation of energy in physics. We should not ignore it. At all points the particle should have the same constant speed - if the speed varies in your simulations it means you have a problem.
2) electric field. Put a sphere of uniform charge in space and now let a particle loose some way off with some speed not in the direction of the charge.
You can read about the sphere of uniform charge here:
http://hyperphysics.phy-astr.gsu.edu/HB ... ph.html#c3
Here are the equations for it:
In the end you should see something like this:
And again we come to this energy thing. Particle energy + potential energy should be const.
So V(mid)-V(particle)+eV(speed(particle))=const throughout the paths. If not you have a problem.
These are just two tests, you should have more, and you should run them regularly to make sure nothing has broken.
And then I'd love to see recirculation from your model. A particle navigating the polywell. Energy not gained and not lost.
After you get all these things working, you might start thinking about doing weird things using trigonometric functions. I know these test are kind of boring, but without those you don't get results, you get fantasy.
Now with ephi I did all these things at first to veryify that everything worked before I even started putting electrons into a complex polywell system. Lets be more professional and methodical.
- Indrek
If none of you people have ever done software engineering before, I'd like to explain a few things: you test it you test it you test it. And with the most simple things at first.
So in this sense this stuff has 0 credibility in my eyes right now. Either using today's horoscope or MSimon to interpret these images doesn't make that much qualitative difference. And by going out there and beating the drum about this you're discrediting yourself and everyone else posting in this forum.
But all is not lost. So not to be totally negative. Here are couple of suggestions for two very simple tests of your code:
1) magnetic field. Put a ring of current on the x-y plane, and inject an electron on the same plane some way off. What you should see is something like this:
As polywell is made up of rings of current it's a good test. And another thing. We have this wonderful thing called conservation of energy in physics. We should not ignore it. At all points the particle should have the same constant speed - if the speed varies in your simulations it means you have a problem.
2) electric field. Put a sphere of uniform charge in space and now let a particle loose some way off with some speed not in the direction of the charge.
You can read about the sphere of uniform charge here:
http://hyperphysics.phy-astr.gsu.edu/HB ... ph.html#c3
Here are the equations for it:
In the end you should see something like this:
And again we come to this energy thing. Particle energy + potential energy should be const.
So V(mid)-V(particle)+eV(speed(particle))=const throughout the paths. If not you have a problem.
These are just two tests, you should have more, and you should run them regularly to make sure nothing has broken.
And then I'd love to see recirculation from your model. A particle navigating the polywell. Energy not gained and not lost.
After you get all these things working, you might start thinking about doing weird things using trigonometric functions. I know these test are kind of boring, but without those you don't get results, you get fantasy.
Now with ephi I did all these things at first to veryify that everything worked before I even started putting electrons into a complex polywell system. Lets be more professional and methodical.
- Indrek
Hello Indrek,
May I ask how you know that Dr. Mike didn't already run unit tests on his sim? I mean, quality assurance is important, but it may already be happening ...
And no, I do not know if Dr. Mike has done the QA or not. Such questions as you pose can (and should) be asked. Respectfully if possible.
Regards,
Tony Barry
May I ask how you know that Dr. Mike didn't already run unit tests on his sim? I mean, quality assurance is important, but it may already be happening ...
And no, I do not know if Dr. Mike has done the QA or not. Such questions as you pose can (and should) be asked. Respectfully if possible.
Regards,
Tony Barry
I don't know, but I've had no indications. And I didn't mean to be disrespectful (although granted I almost always sound like, but don't mind that I really mean good). It's just I've been through this myself and I've got various fantastic results before (as errors) so I have lots of scepticism as I know what can go wrong.
Now I got this book "Field Solutions on Computers" by Stanley Humphries, Jr. (http://www.fieldp.com/book.html) and in the very first chapter you can read some guidelines by the author on how to live your life etc., and here's one: "Never believe anyone over thirty whose presentation hinges on a computer-generated movie". Based on my experiences so far this is excellent advice.
- Indrek
Now I got this book "Field Solutions on Computers" by Stanley Humphries, Jr. (http://www.fieldp.com/book.html) and in the very first chapter you can read some guidelines by the author on how to live your life etc., and here's one: "Never believe anyone over thirty whose presentation hinges on a computer-generated movie". Based on my experiences so far this is excellent advice.
- Indrek
Indrek, I totally agree with you! But the physics of the simulation follow my math notes really well - in other words I'm checking the code against theory. I didn't publish my notes on the static well because they indicated exponential growth and no stable orbit - which didn't look too good. The simulation showed the same thing.
Now, the fundamental difference between what I did before and what I showed this time (and again, I agree with you, we gotta check and tripple check everything since I make mistakes often enough) is the oscillation on the grid. You point out energy conservation - and indeed, the particle should have no net change in velocity in the ABSENCE of an electric field. But the key to the polywell is the E field on the MaGrid which acclerates the electrons into the center to create the well. By using an oscillating field instead the average change in velocity is kept near zero, and in fact you can cool particles if collisions are included (stochastic cooling is done in particle beam lines to improve the beam dynamics - and we're talking about GeV particles!)
The code is posted on http://www.eskimo.com/~eresrch/Fusion/*.c, both field generation and particle tracking. It is trivially simple code. I WANT people to check it!
Simon: I thought about AM radio stations this morning. Glad to know I was on the right track! A 100kW station has got to have some nice tubes on the output stage...
I gotta deal with kids today, but next step is fluid analysis, then full plasma (which I expect to have some trouble with, so Indrek I'm gonna count on your help!)
Full trapping really helps with efficiency, but even if it is only 80% trapping it is a hell of a lot better than 8%.
Now, the fundamental difference between what I did before and what I showed this time (and again, I agree with you, we gotta check and tripple check everything since I make mistakes often enough) is the oscillation on the grid. You point out energy conservation - and indeed, the particle should have no net change in velocity in the ABSENCE of an electric field. But the key to the polywell is the E field on the MaGrid which acclerates the electrons into the center to create the well. By using an oscillating field instead the average change in velocity is kept near zero, and in fact you can cool particles if collisions are included (stochastic cooling is done in particle beam lines to improve the beam dynamics - and we're talking about GeV particles!)
The code is posted on http://www.eskimo.com/~eresrch/Fusion/*.c, both field generation and particle tracking. It is trivially simple code. I WANT people to check it!
Simon: I thought about AM radio stations this morning. Glad to know I was on the right track! A 100kW station has got to have some nice tubes on the output stage...
I gotta deal with kids today, but next step is fluid analysis, then full plasma (which I expect to have some trouble with, so Indrek I'm gonna count on your help!)
Full trapping really helps with efficiency, but even if it is only 80% trapping it is a hell of a lot better than 8%.
Indrek,
There is no doubt I am prone to let my enthusiasms run away with me. It is one of the things that got me started with Polywell. I might add that it is one of the things that got me started with computers in Jan. of '75. That worked out pretty well. At the time the big boys were saying that micro-chips were toys and would never amount to much. Heh.
My model is this - get the word out when something interesting comes up and put more eyeballs on it.
It has worked pretty well so far in moving the open source Bussard reactor along.
Dr. Mike,
Have you tried AC with a DC bias? i.e. 0 - 50 KV vs +/- 50 KV? We have to keep the ions confined after all. Or maybe 50 KV +/- 10 KV.
It would also be useful to know what frequencies are involved.
There is no doubt I am prone to let my enthusiasms run away with me. It is one of the things that got me started with Polywell. I might add that it is one of the things that got me started with computers in Jan. of '75. That worked out pretty well. At the time the big boys were saying that micro-chips were toys and would never amount to much. Heh.
My model is this - get the word out when something interesting comes up and put more eyeballs on it.
It has worked pretty well so far in moving the open source Bussard reactor along.
Dr. Mike,
Have you tried AC with a DC bias? i.e. 0 - 50 KV vs +/- 50 KV? We have to keep the ions confined after all. Or maybe 50 KV +/- 10 KV.
It would also be useful to know what frequencies are involved.
Engineering is the art of making what you want from what you can get at a profit.
Exponential growth? That sphere of charge before is pretty much a static well. There's no exponential growth anywhere. What do you mean? What are you talking about?drmike wrote:Indrek, I totally agree with you! But the physics of the simulation follow my math notes really well - in other words I'm checking the code against theory. I didn't publish my notes on the static well because they indicated exponential growth and no stable orbit - which didn't look too good. The simulation showed the same thing.
The sad reality is that probably no-one will (seriously) look at your code. Also having people look at your code will not bring it any credibility or correctness. But showing that your code works for "standard" situations will do that.drmike wrote: The code is posted on http://www.eskimo.com/~eresrch/Fusion/*.c, both field generation and particle tracking. It is trivially simple code. I WANT people to check it!
- Indrek
Sorry for being dense -- what's the frequency here? I would think that you'd want a wavelength at < size of polywell, but, intuitively I would think that an AC field would remove straight paths for escaping electrons (and in the presence of a mag field, that could prove useful). I'm concerned what it would mean once you add the ions, and I agree with Indrek -- you probably want to validate the code against known conditions first. Unit test your code!drmike wrote:Well, I took a giant step back and found a lot of stupid ideas
But it led me to an idea which I haven't seen bandied about, so I tried one that looks pretty good. I simply changed the E field on the MaGrid from static to be cos(w*t).
Very interesting, though.
/delurk
@ Indrek: Yes, if I want to be "scientific" I have to do a full reality check. I'm pretty happy that my plots (not shown on the web) of particles with no E field but in a mag field look exactly like your plots of particles in a mag field but no E field. That says it is close enough for basement playing. It is not publishable for certain!
@dnavas: I have to work out the frequency. The formula for dimensionless time is
nu = e*mu0*I/(4*pi*m*L) * time
The frequency I used was 5*nu = 2*pi*f*t. That says the frequency is
f = 5*nu/2*pi*time = 5/8 * e*mu0*I/(pi^2*m*L)
I had I = 200k Amp-turns, so f*L = 2.8e9. So if L = 1 meter, it would be 2.8 GHz.
That'll be tough to do, but at L = 10 meter, f = 280 MHz and it is a lot more reasonable.
@Indrek again: The staic well is a problem. An electron can circulate and pick up energy - it does not have to cross the well going up and down hill. It can continuously go down hill until its energy exceeds the containment voltage. The basic force on the charged particles is q*(E + v X B) = m*a. So v X B changes direction and helps containment, but the E causes the energy to increase. If a particle spends all it's time circulating, it will grow in energy. I picked an assumption that caused exponential growth, I think it was a maxwellian particle distribution and energy distribution. So that's why I got exponential velocity increase. But I need to revisit that and double check it with the ion density included (and with the 1/r^2 distribution mentioned by Bussard instead of what I picked).
Time for the messy stuff...
@dnavas: I have to work out the frequency. The formula for dimensionless time is
nu = e*mu0*I/(4*pi*m*L) * time
The frequency I used was 5*nu = 2*pi*f*t. That says the frequency is
f = 5*nu/2*pi*time = 5/8 * e*mu0*I/(pi^2*m*L)
I had I = 200k Amp-turns, so f*L = 2.8e9. So if L = 1 meter, it would be 2.8 GHz.
That'll be tough to do, but at L = 10 meter, f = 280 MHz and it is a lot more reasonable.
@Indrek again: The staic well is a problem. An electron can circulate and pick up energy - it does not have to cross the well going up and down hill. It can continuously go down hill until its energy exceeds the containment voltage. The basic force on the charged particles is q*(E + v X B) = m*a. So v X B changes direction and helps containment, but the E causes the energy to increase. If a particle spends all it's time circulating, it will grow in energy. I picked an assumption that caused exponential growth, I think it was a maxwellian particle distribution and energy distribution. So that's why I got exponential velocity increase. But I need to revisit that and double check it with the ion density included (and with the 1/r^2 distribution mentioned by Bussard instead of what I picked).
Time for the messy stuff...
To me this sounds like you invented the perpetum mobile, energy out of nothing. In my basement the laws of termodynamics are obeyed.drmike wrote:@Indrek again: The staic well is a problem. An electron can circulate and pick up energy - it does not have to cross the well going up and down hill. It can continuously go down hill until its energy exceeds the containment voltage. The basic force on the charged particles is q*(E + v X B) = m*a. So v X B changes direction and helps containment, but the E causes the energy to increase. If a particle spends all it's time circulating, it will grow in energy. I picked an assumption that caused exponential growth, I think it was a maxwellian particle distribution and energy distribution. So that's why I got exponential velocity increase. But I need to revisit that and double check it with the ion density included (and with the 1/r^2 distribution mentioned by Bussard instead of what I picked).
- Indrek
Dr. Mike,
280 MHz in a 10m device will be interesting. The wavelength is 1m so phasing will have to be worked out. In addition you have the problem that the grid will not have a single phase because of its size relative to wavelength.
You have an antenna structure.
On top of that there are standing waves not just on the grid but also inside the device.
The computation of all that is going to chew up a lot of machine cycles. Heh.
I think maybe the POPS idea of "cooling" the ions has a better chance of working. Will the cool (monoenergetic) ions cool the electrons?
The thing to do IMO is focus on getting the fields right then getting ions and electrons into the system followed by trying some POPS stuff.
POPS frequencies should run about 1/60th of the frequencies you show for D-D. About 4 MHz for the 10m case. at 4 MHz the wavelength is 75 m. Rule of thumb is that 1/10th wavelength can be treated as a "DC" case.
If we POPS B11 it should be quite tractable. D-D is on the edge (computationally).
280 MHz in a 10m device will be interesting. The wavelength is 1m so phasing will have to be worked out. In addition you have the problem that the grid will not have a single phase because of its size relative to wavelength.
You have an antenna structure.
On top of that there are standing waves not just on the grid but also inside the device.
The computation of all that is going to chew up a lot of machine cycles. Heh.
I think maybe the POPS idea of "cooling" the ions has a better chance of working. Will the cool (monoenergetic) ions cool the electrons?
The thing to do IMO is focus on getting the fields right then getting ions and electrons into the system followed by trying some POPS stuff.
POPS frequencies should run about 1/60th of the frequencies you show for D-D. About 4 MHz for the 10m case. at 4 MHz the wavelength is 75 m. Rule of thumb is that 1/10th wavelength can be treated as a "DC" case.
If we POPS B11 it should be quite tractable. D-D is on the edge (computationally).
Engineering is the art of making what you want from what you can get at a profit.
I just pulled a number out of my butt and tried it!! There is nothing here but an IDEA.
A lot of physics has to be understood yet. The basic idea is that we can improve electron confinement tremendously by being dynamic rather than static. It is just another idea, the code example shows the idea is reasonable - that is all!! The operating frequency can be swept from DC up to terrahertz in a model and we can find the best confinement for the most reasonable cost and power boost. Then we have to build an experiment and try it!!
In one of Bussard's papers he says "the power balance in Polywell(tm) systems is determined entirely by the rate of electron losses; if these are large, then the system can not yield net power." (see Phenomenological Modelling of Polywell/SCIF Multi-cusp Inertial-Electrostatic Confinement Systems, EMC2-1191-02)
If the power to oscillate the grid is 1/5th or less of the net power gain it is a huge win in terms of building a power plant. I think it will be much less than that.
@Indrek - I have clearly failed to explain things very well. I have been involved with the "free energy" community for the past year or so and it is quite amusing. If I have accidently used their terminology it is a mistake. The point I'm trying to make is that the E field for a MaGrid is zero in the center, high near the grid and falls off toward the outer shell. It is everywhere (excpet at the exact center) positive in terms of the acceleration it adds to an electron. The increasing energy on the electron comes from the power in the MaGrid, but for computational simplicity I just ignore the transfer. The voltage on the grid is considered independent of the load. Even when I do modelling on a plasma I'm going to make that assumption to start with. By using an oscillating MaGrid, the average E field is zero and the electron should not pick up any net gain in velocity, it should just spin. Ideally, the electron actually loses energy and stays cold in the center allowing ions to see a net large accelerating field.
@Simon - If we can keep the electrons cold and not moving much (or on average looking like that) and the ions with high radial energy but little transverse energy, then the system is ideal for power production. Rather than pulsed like POPS, I am thinking in terms of continuous dynamics. Ion acoustic waves in a cold electron plasma with radial motion. It will be fun to play with the physics to see - it may end up looking like POPS anyway.
A lot of physics has to be understood yet. The basic idea is that we can improve electron confinement tremendously by being dynamic rather than static. It is just another idea, the code example shows the idea is reasonable - that is all!! The operating frequency can be swept from DC up to terrahertz in a model and we can find the best confinement for the most reasonable cost and power boost. Then we have to build an experiment and try it!!
In one of Bussard's papers he says "the power balance in Polywell(tm) systems is determined entirely by the rate of electron losses; if these are large, then the system can not yield net power." (see Phenomenological Modelling of Polywell/SCIF Multi-cusp Inertial-Electrostatic Confinement Systems, EMC2-1191-02)
If the power to oscillate the grid is 1/5th or less of the net power gain it is a huge win in terms of building a power plant. I think it will be much less than that.
@Indrek - I have clearly failed to explain things very well. I have been involved with the "free energy" community for the past year or so and it is quite amusing. If I have accidently used their terminology it is a mistake. The point I'm trying to make is that the E field for a MaGrid is zero in the center, high near the grid and falls off toward the outer shell. It is everywhere (excpet at the exact center) positive in terms of the acceleration it adds to an electron. The increasing energy on the electron comes from the power in the MaGrid, but for computational simplicity I just ignore the transfer. The voltage on the grid is considered independent of the load. Even when I do modelling on a plasma I'm going to make that assumption to start with. By using an oscillating MaGrid, the average E field is zero and the electron should not pick up any net gain in velocity, it should just spin. Ideally, the electron actually loses energy and stays cold in the center allowing ions to see a net large accelerating field.
@Simon - If we can keep the electrons cold and not moving much (or on average looking like that) and the ions with high radial energy but little transverse energy, then the system is ideal for power production. Rather than pulsed like POPS, I am thinking in terms of continuous dynamics. Ion acoustic waves in a cold electron plasma with radial motion. It will be fun to play with the physics to see - it may end up looking like POPS anyway.
"the power in the MaGrid," - what's that? If you have only static electric field and static magnetic field - then electrons will not get infinite energy. It's the same as combining those two tests I described before - it will just oscillate around with kinetic+potential(electric)=constant.drmike wrote:@Indrek - I have clearly failed to explain things very well. I have been involved with the "free energy" community for the past year or so and it is quite amusing. If I have accidently used their terminology it is a mistake. The point I'm trying to make is that the E field for a MaGrid is zero in the center, high near the grid and falls off toward the outer shell. It is everywhere (excpet at the exact center) positive in terms of the acceleration it adds to an electron. The increasing energy on the electron comes from the power in the MaGrid, but for computational simplicity I just ignore the transfer. The voltage on the grid is considered independent of the load. Even when I do modelling on a plasma I'm going to make that assumption to start with. By using an oscillating MaGrid, the average E field is zero and the electron should not pick up any net gain in velocity, it should just spin. Ideally, the electron actually loses energy and stays cold in the center allowing ions to see a net large accelerating field.
Also electrons gaining speed is not the problem with polywell - it's the opposite - we want them not to slow down but to go as fast as possible. The potential well they make is directly dependant on their speed. The faster they go the deeper the well - the more fusion we get.
- Indrek
Dr. Mike.
I'm actually thinking in terms of multiple frequencies as a possibility. One for ion acceleration. Another for electron cooling. It is still not clear in my mind how they should be applied. Multiple grids?
I have been thinking about that ever since I thought about POPS and pB11. The protons and Boron POP at different frequencies.
I'm actually thinking in terms of multiple frequencies as a possibility. One for ion acceleration. Another for electron cooling. It is still not clear in my mind how they should be applied. Multiple grids?
I have been thinking about that ever since I thought about POPS and pB11. The protons and Boron POP at different frequencies.
Engineering is the art of making what you want from what you can get at a profit.